Why does statistics seem like sorcery?

PerseusOmega9 · 01-28-2016, 12:57 AM

And I even have an undergrad in math

~~Hobo Brett~~ · 01-28-2016, 12:59 AM

cause it is

PerseusOmega9 · 01-28-2016, 01:03 AM

Well that just confirms my suspicions

Brother Daniel · 01-28-2016, 01:15 AM

I agree with Bart

buttershug · 01-28-2016, 01:20 AM

:

cause it is

How else could Littlewood provide a miracle a month?

gandalf · 01-28-2016, 09:52 AM

Its tru no1z been able 2 xplain margin of error coz it just rose from a pot of boiled newt and frog headz

Pingu · 01-28-2016, 10:29 AM

It's based on the same principle.

In sorcery, you take a little piece of someone, nail clippings, hair, whatever and do something to it, and that same thing happens to the whole person.

Same with statistics.

You take a little piece of a population, a sample, and find something happening in it, and then that same thing happens in the whole population.

PerseusOmega9 · 01-28-2016, 12:57 PM

And one goes into a black cauldron, the other into a blackbox

Mike PSS · 01-28-2016, 01:32 PM

What's not to love?

el jefe · 01-28-2016, 02:55 PM

without knowing what in particular you mean, or much about the subject in general, I am guessing the answer is that a lot of the information statistics allows you to "extract" is actually prepackaged in the model you're assuming. what you're doing is mostly unpacking a bunch of problem-independent information, and only using the problem-specific data to fit a few parameters.

if you assume some set of data follow a normal distribution, e.g., you've just said a lot about it. the real "magic" lies in the fact that some well-studied distributions happen to be good models of some important problems.

el jefe · 01-28-2016, 02:56 PM

and that is how we know that the finns and the norse got it on and had odin

Pingu · 01-28-2016, 03:06 PM

:

without knowing what in particular you mean, or much about the subject in general, I am guessing the answer is that a lot of the information statistics allows you to "extract" is actually prepackaged in the model you're assuming. what you're doing is mostly unpacking a bunch of problem-independent information, and only using the problem-specific data to fit a few parameters.

if you assume some set of data follow a normal distribution, e.g., you've just said a lot about it. the real "magic" lies in the fact that some well-studied distributions happen to be good models of some important problems.

I don't think the biggest problem lies primarily in the "normal distribution" assumption. We can deal with that. I think it lies in the "random sample" assumption.

The law of big numbers does seem like sorcery, even though it is true.

But it's only true if you have a random sample, and in the social sciences, in which statistics are most used and abused, we virtually never do.

el jefe · 01-28-2016, 03:11 PM

maybe we're misunderstanding each other. I took "sorcery" as saying statistics was unreasonably effective, and I was trying to account for a good thing.

Pingu · 01-28-2016, 03:18 PM

Oh, right.

In that sense, the law of large numbers really is like sorcery. And using a bootstrap even more so.

But it's also fake because the p values assume something you never have (a random sample) and you are usually testing a hypothesis that is almost certainly false (the null).

PerseusOmega9 · 01-28-2016, 04:32 PM

I meant by sorcery that I have no idea how it works or why it works, math on the other hand was ways easy

gandalf · 01-28-2016, 05:18 PM

One tiem this scientist f0und a pattern then he did sum xperimentz and found the cauze of teh pattern n0w everytiem we see a pattern there must be sum c0nnection see how it werkz

Mike PSS · 01-28-2016, 05:25 PM

:

I meant by sorcery that I have no idea how it works or why it works, math on the other hand was ways easy

"There are three kinds of lies: lies, damned lies, and statistics."

PerseusOmega9 · 01-28-2016, 05:27 PM

:

One tiem this scientist f0und a pattern then he did sum xperimentz and found the cauze of teh pattern n0w everytiem we see a pattern there must be sum c0nnection see how it werkz

So we can be statistically confident that you're retarded.

F X · 01-28-2016, 05:33 PM

Whatever you do, don't look into mathematical probability. It's far worse.

gandalf · 01-28-2016, 05:54 PM

:

One tiem this scientist f0und a pattern then he did sum xperimentz and found the cauze of teh pattern n0w everytiem we see a pattern there must be sum c0nnection see how it werkz

So we can be statistically confident that you're retarded.

So much h8

Pingu · 01-28-2016, 10:09 PM

:

Whatever you do, don't look into mathematical probability. It's far worse.

Worse than what?

Testycalibrated · 01-28-2016, 11:18 PM

:

I meant by sorcery that I have no idea how it works or why it works, math on the other hand was ways easy

"There are three kinds of lies: lies, damned lies, and statistics."

I prefer, "statistics are the shortest path between an unwarranted assumption and a foregone conclusion."

F X · 02-05-2016, 03:46 PM

Pingu · 02-07-2016, 03:19 PM

Depends what you mean by correlation.

A correlation coefficient can imply causation, if one of the variables was experimentally manipulated.

buttershug · 02-07-2016, 03:36 PM

:

Depends what you mean by correlation.

A correlation coefficient can imply causation, if one of the variables was experimentally manipulated.

I think some people conflate "implies" and "indicates".

01-28-2016, 12:57 AM	#2609490 / #1
PerseusOmega9 Saves People From Themselves : Jul 2010 : Piedmont, NC : 8,861	Why does statistics seem like sorcery? And I even have an undergrad in math __________________

01-28-2016, 01:03 AM	#2609495 / #3
PerseusOmega9 Saves People From Themselves : Jul 2010 : Piedmont, NC : 8,861	Well that just confirms my suspicions __________________

01-28-2016, 01:15 AM	#2609500 / #4
Brother Daniel predisposition to antagonism Kodos the Executioner Mod: HASH, Philo, Math, THC : Jun 2008 : on the shore of the North Atlantic : 15,670	I agree with Bart __________________ Requiem: Everything we humans do is fully deserving of ridicule and mockery. Without a God to laugh, it falls to us enlightened fools to make up the difference.

01-28-2016, 09:52 AM	#2609577 / #6
gandalf teh r3t4rd teh pplz champi0n : Feb 2010 : 13,784	Its tru no1z been able 2 xplain margin of error coz it just rose from a pot of boiled newt and frog headz __________________ el guapo: gandalf, you are god's gift to mafia; i can be a little aspie rmacfarl: Febble you should listen to Gandalf DeanM: Yes I acknowledge you were right. laughing dog: gandalf has it right rednoise: gandalf is right. AwfulCrawler: I don't blame you for being a complete, utter retard gandalf. You were probably born that way. Mattshizzle: Advice # 1 - post in English instead of Retarded. Schneibster: Do you have any idea how many people have arrogantly told me I was full of shit on the Internet?

01-28-2016, 10:29 AM	#2609584 / #7
Pingu I did. F. Poste. GLaDOS : Mar 2008 : UK : 60,846	It's based on the same principle. In sorcery, you take a little piece of someone, nail clippings, hair, whatever and do something to it, and that same thing happens to the whole person. Same with statistics. You take a little piece of a population, a sample, and find something happening in it, and then that same thing happens in the whole population. __________________ Join me at The Skeptical Zone

01-28-2016, 12:59 AM	#2609492 / #2
~~Hobo Brett~~ Banned : Dec 2015 : 268	cause it is

01-28-2016, 12:57 PM	#2609631 / #8
PerseusOmega9 Saves People From Themselves : Jul 2010 : Piedmont, NC : 8,861	And one goes into a black cauldron, the other into a blackbox __________________

01-28-2016, 01:32 PM	#2609666 / #9
Mike PSS Senior Member : Mar 2008 : Florida : 11,478	What's not to love?

01-28-2016, 02:55 PM	#2609738 / #10
el jefe Senior Member TR Pundit : Oct 2008 : 30,945	without knowing what in particular you mean, or much about the subject in general, I am guessing the answer is that a lot of the information statistics allows you to "extract" is actually prepackaged in the model you're assuming. what you're doing is mostly unpacking a bunch of problem-independent information, and only using the problem-specific data to fit a few parameters. if you assume some set of data follow a normal distribution, e.g., you've just said a lot about it. the real "magic" lies in the fact that some well-studied distributions happen to be good models of some important problems. Last edited by el jefe; 01-28-2016 at 02:58 PM.

01-28-2016, 02:56 PM	#2609739 / #11
el jefe Senior Member TR Pundit : Oct 2008 : 30,945	and that is how we know that the finns and the norse got it on and had odin

01-28-2016, 03:11 PM	#2609744 / #13
el jefe Senior Member TR Pundit : Oct 2008 : 30,945	maybe we're misunderstanding each other. I took "sorcery" as saying statistics was unreasonably effective, and I was trying to account for a good thing.

01-28-2016, 03:18 PM	#2609750 / #14
Pingu I did. F. Poste. GLaDOS : Mar 2008 : UK : 60,846	Oh, right. In that sense, the law of large numbers really is like sorcery. And using a bootstrap even more so. But it's also fake because the p values assume something you never have (a random sample) and you are usually testing a hypothesis that is almost certainly false (the null). __________________ Join me at The Skeptical Zone

01-28-2016, 04:32 PM	#2609779 / #15
PerseusOmega9 Saves People From Themselves : Jul 2010 : Piedmont, NC : 8,861	I meant by sorcery that I have no idea how it works or why it works, math on the other hand was ways easy __________________

01-28-2016, 05:18 PM	#2609821 / #16
gandalf teh r3t4rd teh pplz champi0n : Feb 2010 : 13,784	One tiem this scientist f0und a pattern then he did sum xperimentz and found the cauze of teh pattern n0w everytiem we see a pattern there must be sum c0nnection see how it werkz __________________ el guapo: gandalf, you are god's gift to mafia; i can be a little aspie rmacfarl: Febble you should listen to Gandalf DeanM: Yes I acknowledge you were right. laughing dog: gandalf has it right rednoise: gandalf is right. AwfulCrawler: I don't blame you for being a complete, utter retard gandalf. You were probably born that way. Mattshizzle: Advice # 1 - post in English instead of Retarded. Schneibster: Do you have any idea how many people have arrogantly told me I was full of shit on the Internet?

01-28-2016, 05:33 PM	#2609831 / #19
F X Too hot to handle : May 2009 : In the real world : 11,190	Whatever you do, don't look into mathematical probability. It's far worse. __________________ This page archived March 26 2022 by F X

02-05-2016, 03:46 PM	#2612823 / #23
F X Too hot to handle : May 2009 : In the real world : 11,190	__________________ This page archived March 26 2022 by F X

02-07-2016, 03:19 PM	#2613471 / #24
Pingu I did. F. Poste. GLaDOS : Mar 2008 : UK : 60,846	Depends what you mean by correlation. A correlation coefficient can imply causation, if one of the variables was experimentally manipulated. __________________ Join me at The Skeptical Zone

X vBulletin 3.8.6 Debug Information
Page Generation 0.12617 seconds Memory Usage 4,445KB Queries Executed 72 (?)
More Information
Template Usage: (1)SHOWTHREAD (1)ad_footer_end (1)ad_footer_start (1)ad_header_end (1)ad_header_logo (1)ad_navbar_below (1)ad_showthread_beforeqr (1)ad_showthread_firstpost (1)ad_showthread_firstpost_sig (1)ad_showthread_firstpost_start (25)add_ignore_user_to_postbit (10)bbcode_quote (1)footer (1)forumjump (1)forumrules (1)gobutton (1)header (1)headinclude (121)mysmilies_imagebit (1)navbar (3)navbar_link (1)navbar_mini (1)navbar_noticebit (55)option (1)pagenav (1)pagenav_curpage (1)pagenav_pagelink (25)postbit_legacy (25)postbit_onlinestatus (25)postbit_reputation (25)postbit_wrapper (1)spacer_close (1)spacer_open (1)tagbit_wrapper (13)v3arcade_award_bit (3)v3arcade_postbit_userid_popup_menu (3)v3arcade_postbit_userid_trophy Phrase Groups Available: global inlinemod postbit posting reputationlevel showthread	Included Files: ./showthread.php ./global.php ./includes/init.php ./includes/class_core.php ./includes/config.php ./includes/functions.php ./includes/class_hook.php ./includes/functions_notice.php ./mobiquo/smartbanner.php ./mobiquo/smartbanner/head.inc.php ./includes/functions_bigthree.php ./includes/class_postbit.php ./includes/class_bbcode.php ./includes/functions_reputation.php Hooks Called: init_startup cache_permissions fetch_threadinfo_query fetch_threadinfo fetch_foruminfo style_fetch cache_templates global_start parse_templates fetch_musername notices_check_start notices_noticebit global_setup_complete showthread_start showthread_getinfo forumjump showthread_post_start showthread_query_postids showthread_query bbcode_fetch_tags bbcode_create showthread_postbit_create postbit_factory postbit_display_start reputation_image bbcode_parse_start postbit_imicons fetch_userinfo_query fetch_userinfo bbcode_parse_complete_precache bbcode_parse_complete postbit_display_complete pagenav_page pagenav_complete tag_fetchbit_complete forumrules navbits navbits_complete showthread_complete
Messages: